There you have it! BREAKING NEWS! In a single week, the entire economics of AI just got rewritten.
July 9, 2026 · 0 likes · 0 comments
China Threat AI Cybersecurity Workforce
There you have it! BREAKING NEWS! In a single week, the entire economics of AI just got rewritten.
Wednesday, xAI dropped Grok 4.5. Thursday, OpenAI's GPT-5.6 — Sol, Terra, and Luna — goes public after clearing government review.
Sources:
https://lnkd.in/ebhcPnNB
https://lnkd.in/eZXFUHpB
Two frontier launches. One week. And the story everyone will miss is the price tag.
Grok 4.5: $2 per million input tokens, $6 per million output.
Anthropic's Opus 4.7: $5 in, $25 out.
OpenAI's Sol: $5 in, $30 out.
Musk is calling Grok 4.5 "Opus-class" — comparable capability at a fraction of the cost. His own benchmarks, so take the capability claim with salt until independent developers run it. But the pricing is not a claim. It is published. It is objective. And $2/$6 against $5/$25 is not a rounding error. That is a different procurement math for every enterprise on earth.
Here is what nobody in the "AI hype" crowd wants to admit: these models are not chatbots anymore. Grok 4.5 is built to build Excel models from live web research, generate PowerPoints, run legal analysis, write code — the actual desk work millions of people get paid to do. OpenAI's Sol ships an "ultra mode" that spins up subagents to do complex work autonomously.
The tool that replaces the analyst just got cheaper than the analyst's lunch.
That is the whole thesis in one week.
Now the part that deserves an honest look. GPT-5.6 sat in a government review period before this release — held for the Center for AI Standards and Innovation inside Commerce. Voluntary, not a legal mandate. But OpenAI itself said it doesn't want this to become the default, because it "keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them."
I agree with that sentence completely.
We are in a race where China deploys while we deliberate. Voluntary early access to build trust — fine. A permanent gate that slows every frontier release by two weeks while our adversaries ship on their own clock — that is how you lose the lead you already have. The 60-day clock on a formal process is ticking. Whoever writes those rules decides how fast America gets to compete.
Cheaper, faster, more autonomous, and shipping every single week. That is the curve. It does not care whether you are ready.
I wrote a whole book about exactly this. Replacement — out July 28.
Thoughts?
Wednesday, xAI dropped Grok 4.5. Thursday, OpenAI's GPT-5.6 — Sol, Terra, and Luna — goes public after clearing government review.
Sources:
https://lnkd.in/ebhcPnNB
https://lnkd.in/eZXFUHpB
Two frontier launches. One week. And the story everyone will miss is the price tag.
Grok 4.5: $2 per million input tokens, $6 per million output.
Anthropic's Opus 4.7: $5 in, $25 out.
OpenAI's Sol: $5 in, $30 out.
Musk is calling Grok 4.5 "Opus-class" — comparable capability at a fraction of the cost. His own benchmarks, so take the capability claim with salt until independent developers run it. But the pricing is not a claim. It is published. It is objective. And $2/$6 against $5/$25 is not a rounding error. That is a different procurement math for every enterprise on earth.
Here is what nobody in the "AI hype" crowd wants to admit: these models are not chatbots anymore. Grok 4.5 is built to build Excel models from live web research, generate PowerPoints, run legal analysis, write code — the actual desk work millions of people get paid to do. OpenAI's Sol ships an "ultra mode" that spins up subagents to do complex work autonomously.
The tool that replaces the analyst just got cheaper than the analyst's lunch.
That is the whole thesis in one week.
Now the part that deserves an honest look. GPT-5.6 sat in a government review period before this release — held for the Center for AI Standards and Innovation inside Commerce. Voluntary, not a legal mandate. But OpenAI itself said it doesn't want this to become the default, because it "keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them."
I agree with that sentence completely.
We are in a race where China deploys while we deliberate. Voluntary early access to build trust — fine. A permanent gate that slows every frontier release by two weeks while our adversaries ship on their own clock — that is how you lose the lead you already have. The 60-day clock on a formal process is ticking. Whoever writes those rules decides how fast America gets to compete.
Cheaper, faster, more autonomous, and shipping every single week. That is the curve. It does not care whether you are ready.
I wrote a whole book about exactly this. Replacement — out July 28.
Thoughts?